Global Wind Database#
The Wind catalogue connects projects, exact equipment configurations, turbines and
numerical curves. Batch 2026-11-07-2000 contains 8,332 projects, 585 equipment
records, 13,657 linked turbines and two original numerical curves, across
64 countries and 66 equipment manufacturer labels. Merged main at 15da19b4
contains 8,328 projects and the same other entity counts. This run adds three
project records; it does not recount earlier batches. These are research coverage
counts, not a current operating census. Exact continuation queues are in
catalogues/wind/research/latest.json.
All four discovered BOE daily A/B title partitions for July 1–4 are reviewed. Two solar-hybrid corrections are excluded from wind additions and cancellations. Desma’s original 6 May 2025 construction authorization supplies 27.12 MW of wind design capacity, six SG 170-HH135 N6 turbines and 25 MW grid access. Its solar module and shared 165 MW downstream line remain separate. Authorized “potencia instalada” does not become current installed capacity.
Marubeni’s two-project award and JBIC financing disclosures establish Al-Ghat and Waad Al-Shamal project identities, offtake terms and loan boundaries. AXIA’s NTS design tables disclose WD200-7700 fleets, but not exact regional equipment revisions. The 600/616 MW and 500/approximately 516 MW design observations are retained; unresolved planned capacity stays null. Awarded MW remains separate. The original Waad generation unit discrepancy is held without a silent correction. No capacity factors or measured production are inferred from these documents.
POWERCHINA reports all 80 Al-Ghat turbines grid-connected on 10 June 2026, with a reliability trial still ahead. The recurring grid-connection adapter preserves scalar capacity and lifecycle fields, attaches immutable dated observations and rejects incomplete mappings, forecasts and changed source observations. The shared viewer distinguishes first-turbine and full-fleet milestones, unknown counts and missing evidence without adding counts across dates or treating synchronization as commercial operation. The shared JSON exporter now streams bounded batches without changing field order, values or output bytes; temporary output is replaced only after a complete successful write. The BOE adapter also rejects construction/operation claims based only on environmental conditions.
All three new project scopes stay excluded from capacity totals. The 13 historical duplicate-name groups and all 2,216 REPD identity holds remain unresolved; completed REPD partitions were not reprocessed. No exact equipment configurations were fabricated to meet a quota.
The complete 5,344-row Wind partition of WRI Global Power Plant Database v1.3.0 provides a historical discovery baseline. WRI stopped maintenance in early 2022. The initial import enriched two identities with primary evidence; eleven primary-source project records are added, including one aggregate and its two phases. The initial batch contained 5,354 non-group project records. The baseline is not silently classified as current, operational, onshore or free of phase overlaps.
REPD and OEM continuation (3 November 2026)#
The complete 2,858-row REPD Q2 2026 wind partition was reviewed against
WRI’s pinned resources/GBR/matches_GBR.csv crosswalk. 642 existing identities
were substantively enriched (641 operational and one decommissioned observation).
These are dated registry claims published 3 August 2026, not a November operating
census. The other 2,216 applications are deferred, with exact IDs and reasons
in 2026-11-03-0644-repd-decisions.jsonl. No applications were silently added as
new farms, and no identities were merged using a name alone.
All 642 retain their original IDs, creation dates, historical observations and
aggregation_role=unresolved. Their new lifecycle evidence does not resolve
phase or repowering overlap: capacity summaries are unchanged. Applicant/operator
text remains qualified; grid X/Y and generic turbine height remain original values
until CRS and hub/tip definitions are established. The CSV’s capacity field is
interpreted with lifecycle, while CfD values stay distinct in support metadata.
The 22 added OEM records cover all 15 distinct models in the inspected Siemens Gamesa onshore navigation/offshore portfolio, its explicitly announced APAC SG 8.0-167 variant, the B81 blade, and all five models in GE Vernova’s current 3 MW/6 MW tables. B81 compatibility is explicitly sourced to the APAC announcement. India/US variants, flexible modes, frequency-dependent hub/temperature options and derating conditions remain within their documented configurations. Type certification targets are not achieved certificates. GE noise values are retained as qualified observations with scalar acoustics withheld because operating conditions are incomplete. No new curves or individual turbines were added.
The REPD CSV, WRI crosswalk and all 16 OEM HTML captures have original-byte SHA-256 checksums. The inspected Siemens Gamesa August 2026 overview PDF explicitly prohibits reproduction; it is not ingested or redistributed. Independent public product-page tables supply the accepted model facts.
uv run python scripts/stage_wind_repd.py REPD_Publication_Q2_2026.csv matches_GBR.csv \
artifacts/repd-review --retrieved-at 2026-11-03T06:44:32Z
The adapter uses Windows-1252 for this source release, rejects identity collisions,
and records every selected wind entry. It stages enrichment only; it does not
authorize aggregation or overwrite existing primary lifecycle evidence. To
reproduce this iteration, use the pre-iteration catalogue at commit 054d13c.
Storage and reproduction#
Canonical records are UTF-8 JSONL in catalogues/wind/records/, sharded below the
repository source-size limit. UUIDv5 identities use a frozen registry ID or reviewed
natural identity; mutable ratings are excluded. ISO alpha-3 countries, UTC
timestamps, nulls, source registry, field evidence and typed Parquet follow the
PV/BESS architecture. The shared snapshot encoder also preserves the Data Center
public API. Other catalogue source records are unchanged.
uv sync --locked --all-packages --group docs --python 3.13
uv run python -m energy_data_retrieval.wind_catalogue validate
uv run python -m energy_data_retrieval.wind_catalogue build --output artifacts/wind-v1
uv run python examples/database/plot_wind_catalogue.py
Snapshots are immutable, sorted, Zstandard Parquet with schema, source registry, input fingerprint, table hashes, completeness and capacity summaries. Use a fresh output directory. Do not commit generated Parquet; the documentation build recreates it and the lossless viewer projections. Reproducibility is checked with the locked PyArrow version, not promised across arbitrary encoder versions.
The staging adapter reads a locally acquired CSV without network access:
uv run python scripts/stage_wind_gppd.py --help
It processes every primary-fuel Wind row, retains excluded/deferred decisions, rejects duplicate source IDs and preserves observed and modeled generation separately. Synthetic offline fixtures test zero, null, malformed capacities and identity collisions. Canonical rows include curated reconciliation after staging.
Capacity, identity and performance rules#
Boundary |
Representation and aggregation |
|---|---|
Installed, commissioned, planned, awarded, export |
Separate MW fields; never interchangeable or additive |
Historical WRI nameplate |
|
Groups and phases |
Parent is |
Unreviewed identity/scope |
|
Repowering |
Stable generation and |
Mixed fleets |
Explicit manufacturer/model/count entries; unknown counts stay null; linked equipment only when exact |
Count × rating |
A derived consistency check; does not replace a reported capacity |
Generation/resource |
Kind, period, units, reporting boundary, source and locator; resource height required, nullable if unknown |
Capacity factor |
Not computed in this batch: usable capacity-time/partial commissioning boundaries are absent |
Curves |
Separate linked records with original numerical or digitized origin, air density, operating mode and source revision |
Operational evidence is dated, not a guarantee of current operation. Dumat Al Jandal retains its 2019 announced/award evidence, while MacIntyre’s latest reviewed operator update establishes partial operation without a current commissioned MW. Lake Turkana’s installed scalar is withheld pending reconciliation of 310/310.25 MW. WindFloat Atlantic’s 25 MW, Kincardine’s 50 MW and Kallpa’s 344 MW remain reported values despite different count-times-rating checks. Greater Changhua 1 and 2a contribute only their phase capacities; the 900 MW aggregate is excluded.
Equipment and field evidence#
Vestas’ EnVentus, 4 MW and 2 MW portfolio tables and eleven Nordex Delta4000 model pages are processed completely for the selected table/model batches. V236-15.0, the documented V164-8.4 and V164-9.5 configurations, and V47-660/660-200 revision 943111.R4 complete the first equipment batch. Flexible Nordex operating ranges remain qualified options; the malformed V136-4.2 nominal rating is withheld. Hub-height choices do not create synthetic variants.
Every populated scientific field has source IDs plus evidence locator, confidence, original value/unit or documented normalization. Source records retain revision, publication date if available, retrieval date and byte availability. Midnight UTC retrieval timestamps are explicitly date-normalized, not exact access times. The initial WRI CSV checksum covers acquired original bytes. Initial rendered OEM/project pages and the V47 PDF do not claim original-byte checksums; the continuation’s original CSV and OEM HTML captures do. Public access is not a claim of open licensing; only factual specifications are transcribed. The WRI-derived partition is attributed under CC BY 4.0.
Broad schema coverage includes ownership shares, lifecycle dates, grid and offshore infrastructure, resource observations, performance, investment, PPAs, permits, dimensions, electrical/control details, environmental packages and certification scope. Many fields are currently null. Component compatibility requires explicit model-level evidence. Beyond the linked B81 blade, separate gearbox, generator, converter, tower, foundation and floating-platform models remain ranked research tasks; no generic product families are counted as exact equipment records.
Viewer, examples and continued research#
The Wind viewer reuses the existing map clusters, filters, expandable records, source links, CSV export and completeness display. Country, owner, manufacturer, exact model, lifecycle and technology categories and numeric capacities are searchable. Capacity KPIs separate lifecycle states and exclude unresolved overlap scopes. Generic capacity sum charts are disabled for Wind Projects. Equipment statistics describe catalogue specifications, not market share or verified operational performance.
The gallery covers country coverage, installed versus pipeline categories, manufacturer/model distributions, rotor/specific-power and hub options, offshore depth/foundations, and compatible original numerical power curves.
catalogues/wind/research/latest.json is the batch checkpoint;
backlog.json ranks exact remaining sources and offsets. Candidate-level JSONL
ledgers retain additions, within-import enrichment, duplicate candidates, exclusions
and field deferrals. The first USWTDB download failed with HTTP 403 in this
environment: this is an access failure, not evidence of authentication or a reuse
restriction. NREL’s turbine-model archive was inspected but not ingested because
individual curve provenance/conditions require reconciliation.
Hourly continuation must inspect current main, compatible open PRs and the ledger, resume unfinished partitions, acquire a nonblocking branch lock, and publish using an expected-head check. Leave PRs open and unmerged; never overwrite another run.
Administrative-name reconciliation and first converter#
The next bounded review exhausted all 55 deferred changed-name REPD candidates. Only removal of an administrative resubmission label, corroborated by the explicit WRI registry crosswalk and unchanged reported capacity, can pass the opt-in adapter rule. Extension, phase, repowering and revised-application words are retained. Repeated names and replacement links remain held. Capacity aggregation exclusions are preserved even after a successful lifecycle enrichment.
The first exact wind converter cabinet is ABB ACS880-77CC-6880A/5500A-7, from hardware manual 3AXD50000813605 Rev A (22 December 2021), pp87–92. Its grid-side active kW, generator typical kW and apparent kVA remain distinct. Temperature/altitude derating and enclosure-only lifetime conditions are retained; no turbine compatibility or optional-configuration combinations are inferred. The complete ratings table contains one cabinet, not a 20-model equipment batch.
Procedural history continuation#
PR #89 is merged. Main at 33eec838 contains 5,355 projects, 53 equipment records
and two curves. The next batch adds 232 dated planning/appeal events to 114
existing projects without changing lifecycle or capacity eligibility. The first
200 unlinked registry applications retain explicit identity holds and planning
references in the 0835 ledger. No exact 77LC/87LC rating table was obtained;
generic converter ranges and unrelated industrial drives were not imported.
Canadian inventory reconciliation#
The 2026-11-03 09:35 continuation reviews all 327 project-name groups from 7,841 original NRCan workbook rows. It enriches 117 existing Canadian project identities with reported manufacturer/model, rating, rotor diameter, hub height and commissioning groupings; 210 identities remain deferred.
Acceptance requires a unique normalized Canadian project name, all turbine
coordinates within 10 km of the historical project coordinate, project-capacity
agreement within 1% (0.1 MW minimum tolerance), and unique nonempty turbine IDs
within the source group. These are corroboration checks, not proof of phase or
repowering scope. All prior lifecycle fields, capacity values and aggregation
exclusions remain unchanged. Inventory counts and mixed fleets are retained in
extensions_json.nrcan_inventory; current fleet completeness remains unknown.
No individual turbine or exact OEM equipment record is inferred from this review.
The offline wind_catalogue.canada.reconcile_group adapter accepts decoded
source dictionaries. Original workbook units and commissioning strings are
retained. The source registry and decision ledger preserve the original checksum
and Excel-row locators. See nrcan_wind_turbine_database_2026 in the citation
catalogue. The portal update date is not a commercial-operation date.
Swedish registry areas, turbine inventory and VENSYS equipment#
The 10:35 batch covers the complete OMRID < '0500' partition in the official
Vindbrukskollen MapServer: Stockholm, Uppsala and Södermanland. Layers 2 and 3
supply 89 development areas; layers 5–12 supply 348 georeferenced turbine rows.
Seventy-two named area identities and 321 explicit VERKID children are added.
Ten repeated municipality/name areas, five unnamed areas and two areas without
linked turbines remain deferred; two turbine rows lack a published parent area.
All ten pre-existing Swedish records are preserved. Names and proximity to any
existing Swedish project trigger reconciliation rather than automatic addition.
A registry area is a documented development scope, not proof of an independent
physical farm or repowering generation. Every new project retains unresolved
aggregation scope. Uppfört means physically erected: current operation remains
unknown. Permitted, dismantled and withdrawn states map only where explicit;
mixed states and refused applications keep raw history without a guessed current
lifecycle. Source dates, organizations, permit types, modeled annual production,
maximum turbine ratings and boundary polygons remain attached. Reported zero
annual production is preserved; zero dimensions and placeholder model/date
values become null in normalized fields, with originals retained.
Project map anchors use an explicitly identified linked turbine coordinate, not an invented centroid. Turbine points are transformed by the source service to EPSG:4326; positional accuracy remains unquantified. The viewer’s turbine table has parent IDs, model labels, dimensions, filters, clustered map and inventory KPIs. Counts include proposed and historical records. Fleet composition is partial; no inventory label is promoted to an exact OEM configuration.
The reusable offline adapter accepts complete source GeoJSON queries:
uv run python -m energy_data_retrieval.wind_catalogue.sweden \
--input /path/to/downloaded-queries --source-prefix vbk-20261103 \
--timestamp 2026-11-03T10:45:00Z \
--output /tmp/vbk-staged.jsonl --ledger /tmp/vbk-decisions.jsonl
Use the exact ten query URLs and checksums in
research/2026-11-03-1035-source-review.json; save responses as
vbk-layerN.geojson. This staging command never modifies canonical data and
refuses output-file replacement. Existing IDs are held for explicit enrichment
review. To reproduce the accepted first partition, use the pre-batch catalogue
at fc308ef3; rerunning against the updated catalogue correctly holds its IDs.
See vindbrukskollen_open_inventory_2026 in the citation catalogue. Both original
ZIP metadata files identify the inventory as CC0; background maps are separate.
All seven currently linked VENSYS product sheets were reviewed, with original
PDF checksums and p2 locators (vensys_technical_sheets_2025). Tower/IEC and grid
frequency options remain nested within each exact configuration. VENSYS 175
uses a medium-speed geared drivetrain; earlier direct-drive designs are not
silently generalized to it. Thermal derating conditions and acoustic mode
qualifications are retained. The VENSYS 136 PDF and undated webpage disagree on
sound power; the normalized scalar is null pending revision/mode reconciliation.
Numerical AEP tables retain unknown density/distribution and cannot substitute
for instantaneous power curves. Seven plotted curves are deferred, not digitized.
Germany’s MaStR wind XML was acquired via verified HTTP206 ZIP byte ranges:
44,351 unit rows, plus both catalogue dictionaries. Member SHA-256 values are
recorded; a checksum for the undownloaded full archive is explicitly unavailable.
The next task is the complete Schleswig-Holstein named-farm partition after
catalogue-code verification and WRI reconciliation, not automatic turbine
clustering. See mastr_full_export_20261103 and the ranked backlog.
Östergötland and ENERCON continuation#
The next complete partition (0500 <= OMRID < 0600) reviews 227 areas and
672 turbine rows. Independent count queries agree with every downloaded layer.
It adds 192 projects and 559 turbine children; 21 repeated-name scopes, 12 unnamed
areas and two existing identity/spatial candidates remain held (113 turbine rows).
All 5,810 preceding records and all six capacity-category totals remain unchanged.
The adapter now optionally verifies an independent source count, catching silent
truncation even when the response omits an exceeded-transfer-limit flag.
Three ENERCON August 2026 sheets describe four exact configurations: E-138 EP3 E3,
E-160 EP5 E3, E-175 EP5 E1 and E2. The E2 nominal 6,000/7,000 kW options remain
nested; no single rating is selected. Mode/height-dependent acoustic observations
also remain qualified metadata. Cold-climate packages stay distinct from normal
temperature limits, and design-life claims keep their IEC scope. Annual-yield
plots are not numerical power curves. See enercon_technical_sheets_202608.
All three originals downloaded successfully after web-reader errors; this was
not authentication failure or evidence of reuse restrictions.
Next Swedish partition: all Jönköping registry IDs (0600 <= OMRID < 0700).
The German Schleswig-Holstein named-farm reconciliation task remains pending.
Schleswig-Holstein registry continuation#
The complete MaStR Schleswig-Holstein partition contains 4,788 registered units
in 1,739 name groups, including the empty-name group. Conservative operator,
municipality, coordinate, alias and historical-catalogue screens accept 22 named
parks and 105 turbine children. The other 1,717 groups remain in the decision
ledger. Park names are not official farm identifiers. Stable unit registry IDs
anchor future reconciliation; ownership/name changes must not create new farms.
All 22 parks retain unknown farm lifecycle, partial fleet completeness and
unresolved aggregation scope. Unit status, commissioning dates and separate
gross/net kW are retained without substituting sums for reported farm capacity.
The offline wind_catalogue.mastr.stage adapter uses the pinned XML dictionary
and a complete state partition; its synthetic fixture tests alias collisions,
invalid scopes, replay identity and the distinction between unit and farm ratings.
See mastr_full_export_20261103 and the 1235 decision ledger.
The ENERCON upgrade presentation adds exact E-82 E2 and E-103 EP2 model records
with their respective OM-YO-01 and OM02sBV optional modes. Base ratings and
geometry remain unknown because this source does not specify them. Yield claims
retain annual-average wind-speed and site-load-study conditions; they are not
measured operational performance. Graph-only curves remain deferred. The combined
August 2026 overview duplicates the four configurations already catalogued.
See enercon_yield_modes_e82_e103.
Next German partition: Niedersachsen, dictionary code 1408, after verification against the original category 101 dictionary. Reconcile accepted Schleswig-Holstein unit links and held groups separately from new state intake.
Niedersachsen and integrated drivetrain components#
The complete 9,034-unit MaStR state partition (code1408) contains3,046namegroups. The adapter accepts73parks and366linked units;2,973groups remain held. State selection is explicit and dictionary-checked; the default Schleswig-Holstein API remains compatible. Previous records and all capacity summaries stay unchanged.
Winergy’s historical HybridDrive table provides PZFG2456 and PZFG2535, each
rated3000kW, with2000/2500kNm torque and25/31tonnes assembly mass. These are
integrated two-stage gearbox/permanent-magnet generator assemblies, recorded
once per exact model. Original units and normalization are retained. Current
availability, specific turbine compatibility and absolute lifetime remain unknown.
PDF creation metadata2013-06-03 is not asserted as publication date. The peak
efficiency marketing claim lacks operating-point conditions and is not accepted
as normalized performance. See winergy_hybriddrive_pzfg, original PDF p2.
NGC’s independent product page describes broad platforms without an exact model rating table; no artificial configurations were generated from power ranges. Next: Nordrhein-Westfalen (code1409,6,753units), Jönköping and exact component configuration evidence.
Nordrhein-Westfalen and exact yaw/pitch gearboxes#
The complete6,753-unit MaStR partition(code1409) yields67parks and314linked units from2,732namegroups;2,665groups remain held. All prior records and six capacity summaries stay unchanged. Added state support preserves default APIs.
The original NGC regional archive PDF supplies seven exact SYW yaw-drive types
and seven SPT pitch-gearbox types. Rated-torque ranges, peak torque and gear
ratios remain distinct; stage-count options stay within each model. Generic
platform temperature, certification and turbine-compatibility claims are not
inherited. Physical pages9/10(printed08/09) were visually reviewed. Five SMG
platform columns remain held. See ngc_wind_yaw_pitch_2022. The German2024URL
returned404; the regional2022PDF downloaded successfully. This was not an
authentication or explicit reuse restriction. Winergy’s unnamed6MW exhibition
configuration remains held for exact identity. No numerical curves added.
Brandenburg, Jönköping and Bonfiglioli continuation#
Brandenburg’s complete5,334-unit partition yields22parks and183units, with 1,471namegroups deferred. Jönköping’s151areas/902turbines yield109projects and 608units. All ten layers match independent source counts. Forty-two areas and 18orphan units remain held, covering294turbine rows. Both inventories retain unresolved aggregation scope. All7,530preceding rows and capacity totals stay unchanged. Exact query URLs and original response hashes are in the1535review.
Bonfiglioli R05_2 p14 provides nine distinct TW model names across13pitch/yaw
application rows. Each model keeps application-specific peak torque together,
including differing yaw/pitch values. These are OEM reference limits, not nominal
torque or independently validated performance. The19dimension rows on p15 use
T designations and remain held pending T/TW reconciliation. See
bonfiglioli_wind_r05_2. NGC2024original download returned404 while search and
opened reader show different editions; MPPY/MPPB import remains deferred.
Sachsen-Anhalt, Kronoberg and ABB continuation#
Complete state/county partitions yield64projects and433linked units. All8461prior records remain byte-identical; capacity summaries unchanged.878German groups and44Swedish areas (326units) remain held. New scopes are excluded from totals.
The complete15-row ABB2018RevA p10 table (abb_wind_converters_2018_reva)
preserves exact77LC/87LC types, including documented twin cabinets. Typical
generator kW is separate from generator/grid kVA and nominal side currents.
Dimensions, mass and flow retain the original per-cabinet expressions. Ratings
depend on selected options; historical specifications do not establish current
product lifecycle, turbine compatibility or independently validated performance.
Bayern, Kalmar and converter technical conditions#
Complete Bayern1889-unit and Kalmar204-area/1330-unit partitions add168projects and918linked turbines.823German groups,71Swedish areas and24orphan units remain held. All new project scopes remain excluded; prior capacity totals unchanged.
ABB2018RevA p8 explicitly covers77LC/87LC and adds qualified series-level operating/storage temperatures, cooling inlet limits, altitude, electrical interfaces and optional limits to15existing models. All p10ratings remain. Standard altitude1000m is separate from optional4000m; optional coolant55C is not a standard limit. No independent performance/certification claim is added. BMD R04_1 industrial identifiers remain held pending their wind-specific mapping.
Baden-Württemberg, Gotland and ACU control inverters#
Complete partitions add83projects and369turbines; all10059prior records and capacity totals remain unchanged.514German groups,28Swedish areas and3orphan units remain held. New project scopes stay excluded from capacity sums.
Bonfiglioli ACU VEC521R6 pp38–41 (bonfiglioli_acu_vec521r6) adds all18exact
ACU401type columns in sizes1–5. Motor shaft power, output current,60s/1s overload,
2kHz losses, dimensions and approximate mass retain boundaries. ACU401-22
keeps22A main-table and frequency-dependent23A/22.7A observations separately.
WIND R05_2 p13 supports Active Cube series yaw application, not exact turbine
compatibility. No artificial interface/cooling variants or performance claims.
Mecklenburg-Vorpommern, Blekinge and completed ACU type tables#
The 19:35 batch follows the merge of PR #101. From main 2fd691a7, it adds
52 projects and 188 linked turbines across complete 2,741-unit German and
58-area/158-unit Swedish partitions. All 10,529 preceding records remain
unchanged. The 722 German groups and 25 Swedish areas covering 72 unit rows
retain explicit deferral decisions. No new project scope enters capacity totals.
Bonfiglioli VEC521R6 (bonfiglioli_acu_vec521r6) adds all 12 ACU 201 columns
on pages 35–37 and all eight remaining ACU 401 columns on pages 42–43.
The earlier backlog’s size-8 reference was incorrect: page 44 contains derating
diagrams. A separate original document and model-identity review is required.
The records preserve switching-frequency current tables separately from headline current, including ACU 201-19 and -22 discrepancies. ACU 201-15’s single/two-phase 9.5 A limit is distinct from the 12.5 A main-table value. Overloads retain 60 s, 1 s and ten-minute recurrence boundaries. Three-phase-only, choke, thermal, UL scope and conditional IP20 statements remain qualified. WIND R05_2 page 13 supports series-level yaw-drive application; it does not identify an installed turbine interface. The gallery exposes those phase/frequency conditions.
Rheinland-Pfalz, Skåne and ACUx10 frames#
The 20:35 batch follows the merge of PR #106. Complete Rheinland-Pfalz and Skåne inventories add 295 projects and 826 turbines. All 10,789 preceding records and all six capacity totals remain unchanged. The 862 German name groups, 519 Swedish areas and 26 orphan turbines retain explicit holds. The Swedish partition includes prefix11 and confirms no returned rows there.
Bonfiglioli R04_0 (bonfiglioli_acu_r04_0) pages24–25 provide 24 exact
ACUx10 configurations: six frame7 and eighteen frame8 entries. Frame7/8
ACU410-51/-53 identities remain distinct, with different dimensions, mains
currents and peak limits. Frame7 ACU410-53 is liquid-cooled only. Neither
mechanical options nor brake-chopper choices create artificial records.
The six ACU610 voltage-row notes about600V-for-UL remain qualified observations; the approval row supplies no claim. Scalar approval stays unknown. The gallery shows frame/voltage/current boundaries. No ACUx01 replacement, turbine interface or independently validated performance is inferred. The separately discovered spindle application manual was excluded for its explicit reuse restriction.
Hessen, Halland and remaining ACUx10 tables#
The 21:35 continuation follows merged PR #111 and preserves the new gallery hierarchy. Complete Hessen (1,807 units / 627 groups) and Halland (176 areas / 691 units) partitions add 139 projects and 555 linked turbines. All 11,934 preceding records, prior source entries and six capacity totals remain unchanged. All new scopes remain excluded pending phase and repowering reconciliation. Holds cover 589 German groups and 75 Swedish areas plus two orphan units (306 turbine rows). No country or manufacturer is newly counted.
Bonfiglioli R04_0 pages 19–23 (bonfiglioli_acu_r04_0) supply 34 exact
configurations: 12 ACU210 and 22 ACU410 columns. The source checksum was reverified
and all five tables visually reviewed. Input-phase ratings and column-specific
line-choke requirements stay distinct. Standard, coldplate and liquid dimensions
and brake options remain nested. ACU410-12 is explicit on page 21 despite its
omission from the nomenclature summary. Unclaimed approvals remain unknown.
Series yaw applicability is documented; turbine compatibility and independently
validated operational performance are not inferred. The gallery includes the new
230 V ACU210 entries.
Thüringen, Västra Götaland and line chokes#
This batch adds 618 projects, 2,030 turbines and 23 exact LCVS/LCVT line chokes. Merged main at the iteration baseline contains 6,713 projects, 202 equipment, 5,745 turbines and two curves; proposed totals are 7,331 / 225 / 7,775 / 2. All 12,662 preceding records and capacity totals are unchanged. Complete registry partitions were reviewed; 371 German groups and 226 Swedish areas remain held. Thirteen duplicated Swedish turbine IDs quarantine their entire parent scope. Missing area boundaries remain null. All new project scopes are excluded from MW totals.
OEM pp52–56 preserve explicit inverter pairings and original dimension letters. LCVT025 has conflicting inductance/loss rows; scalar values and the disputed ACU410-23 pairing are held. No guessed correction or turbine compatibility is added. The remaining 13 AL-suffix types are the next bounded equipment partition.
Sachsen, Värmland and electrical components#
The 23:35 increment adds 70 projects, 636 linked turbines and 28 exact equipment
types: 13 AL-suffix line chokes and 15 EMC filters. Proposed totals are 7,401
projects, 253 equipment, 8,411 turbines and two curves. Merged main at 7af547f9
contains 6,713 / 202 / 5,745 / 2; PR #119 cumulatively adds 688 / 51 / 2,666 / 0.
All 15,333 records preceding this increment and all capacity totals are unchanged.
The completed Sachsen partition has 482 held groups; Värmland has 44 held areas
covering 734 turbine rows. All 70 accepted project scopes remain excluded from MW totals.
AL-choke current and dissipation depend on the printed inverter application; common inductance and assembly dimensions stay on one exact component record. Two choke-to-inverter links lack a frame identifier and remain held. One filter application prints ACU210-12, which has no exact existing configuration; its label is retained without an inferred alias. EMC class/cable conditions remain qualified, and the general 480V heading is not silently applied to the ACU610 application. Next: Saarland, Örebro and book-filter assembly tables pp62–63.
Saarland, Örebro and braking resistors#
The 4 November increment adds72projects288turbines11exactBRresistor types and enriches12EMCfilters with assembly specifications (R04_0 pp62–65). Saarland114groups andÖrebro25areas/166units remain held. New project scopes remain excluded from MW totals. Two-resistor combinations retain quantity and combination ratings within one component record; no artificial assembly model. HLD110-500/42dimensions and thermal-switch assignments remain unknown. All previous scalar ratings and creation timestamps are preserved.
City states, Västmanland and Dalarna#
The 4 November continuation adds 53 projects and 637 linked turbines, and enriches 11 existing resistor records with legacy VEC501R0 electrical, thermal, mass, terminal and original drawing observations. Current-revision claims are not inferred. The 103 German groups and 47 Swedish areas held for identity review remain in decision ledgers. All new project scopes stay excluded from capacity totals; all previous values and creation timestamps are preserved.
Gävleborg and offshore identity screening#
The 4 November 02:35 increment adds 56 Gävleborg project areas and 928 linked turbines from all 88 areas and 2,124 turbine observations. The remaining 32 areas/1,196 observations are held for identity/scope review. All 1,813 German EEZ units were screened across 48 name groups; aliases and sub-array labels prevent automatic farm additions. The existing E-82 E2 gains revision-qualified 08/2019 base-model specifications. Two restricted Bonfiglioli manuals are excluded. No new project enters capacity totals; all prior specifications are preserved.
Västernorrland continuation#
The 03:35 batch reviews 97 areas and 2,177 linked/orphan turbine observations, adding 55 projects and 847 turbines. Forty-two areas and 117 orphan turbines remain held; no project capacity is inferred. All previous records are unchanged. The reusable quarantine helper now rejects missing IDs before grouping, so missing identities cannot be silently hidden as duplicate conflicts.
Jämtland and converter monitoring#
The 04:35 batch reviews all 138 Jämtland areas and 1,729 turbine observations, adding 69 projects and 784 linked turbines. The remaining 69 areas and 945 turbines are deferred; no project capacity is inferred. ABB system description 3AXD50000022022 Rev C (2020-04-27), pp. 13–14 and 24–25, supplies separately dated system interfaces for 15 existing exact converters and explicitly names the optional NETA-21 monitor. Its record and compatibility links preserve that series/revision boundary; historical ratings are unchanged. Detailed monitor limits remain unknown after an HTTP 403 download failure. Catalogue claims are manufacturer documentation, not independently validated performance.
Västerbotten continuation#
All 130 areas and 2,316 turbine observations were reviewed; 67 projects and 916 linked turbines were accepted. The remaining 63 areas/1,400 turbines remain deferred. Capacity totals are unchanged. NETA-21 gains separately qualified Rev Q (2026-09-18) technical-table observations, pp.225–226; contradictory temperature endpoints are retained without a normalized scalar, and joint NETA/NEXA interface descriptions do not infer per-device allocation.
Norrbotten and monitoring extensions#
All 68 Norrbotten areas and 2,566 turbine observations were reviewed; 37 projects and 822 linked turbines were accepted, with 31 areas/1,744 turbines held. Capacity totals remain unchanged. ABB Rev Q pp.21–24 resolve named NETA-21 connector assignments and explicitly connect the optional NEXA-21 extension. No direct turbine compatibility or joint-table limits are inferred. The Marne E103 appendix is excluded for its explicit reuse restriction.
French network installations and Swedish residual check#
The 07:35 iteration checks all 2,458 EOLIE rows in the official ODRE snapshot of 31 August 2026 against national names/communes and existing French records. The bounded cohort covers installations with industrial-service dates from 2022 and reported installed capacity of at least 1 MW. Thirty-four named onshore installations pass the conservative identity screen. All retain unresolved farm/phase/repowering scope and remain excluded from capacity totals. Installed and maximum grid-injection ratings are separate observations. Network groups are not turbine counts, HTA/HTB are not numeric voltages, and administrative communes are not surveyed project coordinates. Native rolling energy values retain their reporting-boundary limitations; no capacity factors are calculated. See the dated decision ledger for every screened row.
All ten selected Swedish land layers return zero records for null IDs or OMRID >= 2600, completing that partition scan. Deferred identities and offshore coverage still require separate work. USWTDB alternate downloads returned HTTP errors; no authentication or reuse prohibition is inferred.
The complete GE 4.2 MW-117m table adds one exact equipment configuration; its published 4.23 MW nameplate is preserved separately from the model label. Frequency-specific hub options stay nested, and the inadequately qualified noise range is not promoted to a scalar sound-power value. The Vestas 4 MW brochure is excluded under its explicit final-page reuse restriction.
Regulatory decisions and hybrid boundaries#
The BOE adapter reviews Series A notices mentioning wind, including decisions about associated solar plants or auxiliary supplies. It requires complete reviewed selections and checksummed originals before staging records. Its synthetic offline tests reject title-based capacity inference, missing notices, duplicate identities and cancellation of a wind farm from a solar decision.
The Spanish continuation adds five project identities from four complete daily title partitions and two supplemental notices. Final and superseded proposals, withdrawals, conditional environmental declarations and construction permits remain distinct. Conflicting geometry and electrical fields stay null with original observations retained. Unknown operating status is not converted to zero capacity. All new farm/phase scopes remain excluded from capacity totals.
The shared viewer’s expandable metadata exposes boe_procedure and field
sources; the offline Wind example prints decision target, dated wind status,
capacity categories and exclusions. A solar/auxiliary decision cannot be read as
a wind commissioning event. The BOE partitions are not a national wind census.
NEM survey and small-wind certification batch#
Batch 2026-11-04-2135 adds 67 projects and nine exact equipment models.
AEMO’s October 30 workbook supplies the complete wind partition with survey IDs
below 2000: 158 generating-unit rows at 125 survey sites. Sixty-six distinct
sites are added; 59 name/alias/phase candidates are held for explicit crosswalks.
All new scopes remain excluded from capacity totals. Eight explicitly named
phases receive stable group links without additive parent rows. Individual
turbines and exact deployed models remain unknown. NEM coverage excludes
Western Australia and the Northern Territory; NSW1 includes ACT.
Reported unit aggregate AC nameplates are never multiplied by turbine counts. Mixed lifecycle observations remain separated, shared hybrid-site dispatch limits do not become wind capacity, and forecast commercial-use dates do not become COD. The 24–26 September BOE partition adds El Tabaibal (ER150053), whose public-utility modification establishes neither current operation nor a normalized capacity category.
The complete PNNL July 2026 table supports nine exact small-wind models, eight manufacturer/applicant labels, and power at 11 m/s under standard sea-level conditions. These are dated certification observations, not nominal model ratings or independently validated field performance. Acoustic limits at 60 m are kept separate from sound power; unknown rotor geometry precludes specific-power comparisons. No numerical curves were available from this table.
The separate ICC-SWCC directory is excluded because its linked terms explicitly restrict extraction and redistribution. AEMO’s workbook downloaded successfully although its landing page returned HTTP 403. These are distinct access/reuse outcomes. The equipment target of 20 was not met; no unsupported variants were added.
See research/2026-11-04-2135-source-review.json, the AEMO candidate and decision
ledgers, and the PNNL transcription. Next: AEMO survey IDs 2000–2999, then
3000 and above; BOE September 21–23; OEM configuration documents for EOX S-16
and KW20/KW30. Prior reconciliation holds remain queued.
NEM survey IDs 2000–2999 and small-wind OEM specifications#
Batch 2026-11-04-2235 reviews all 111 survey sites in this half-open partition:107 additions and four deferrals. All additions remain excluded from capacity totals. Three explicitly named phases retain stable group IDs, including Goyder South1B linked to the earlier1A family. Twenty sites contain21 generating-block count placeholders; their physical turbine counts stay null, while original counts and reported AC capacities remain reviewable. Wooroora/Chalumbin lifecycle and Clarke Creek/Baldon phase scope remain held.
Five exact equipment models are added:SD3,SD3EX,SD6+,SD12 and the Powersync25 inverter. Existing Excel15 and SD6 identities gain geometry and qualified technical observations without replacing PNNL power at11m/s. SD3EX3/2.5kW observations conflict, so its scalar is null. SD6+ power depends on unspecified inverter settings. Powersync25 factory20kW,table23kW and field-adjusted25kW limits are separate. Nominal and11m/s ratings are excluded from a common nameplate-density comparison. No graph is silently digitized.
Kodair pages were acquired but excluded under explicit republication terms. Eocycle page/PDF returned403; this is a download failure, not an authentication or reuse finding. Complete source decisions, count quarantine IDs, checksums and exact continuation partitions are in the batch research files. The20-equipment target was not met; no variants were invented.
Final NEM partition, local BOE reconciliation and Goldwind tables#
Batch 2026-11-04-2335 adds32projects and15exact equipment models, and enriches one historical project and two existing equipment records. Branch totals are8,168projects (8,167non-group),372equipment,13,633turbines and two numerical curves across59countries. The full AEMO October2026 wind selection is exhausted:269sites/309units,198accepted identities and71deferred. This final33-site partition adds25sites, holds eight overlapping identities, and quarantines twelve count placeholders; all accepted survey scopes remain excluded from capacity totals. BullawahStage2 shares the existing Stage1 family identifier.
The complete September21–23 BOEA/B daily partition yields Ferromolín PEol-1023:27planned MW, six proposed turbine positions, five separate reserve locations and a qualified material-execution budget. A separate2024BOE local inventory within10km of Martelilla yields six additions and one CortijoI historical-record reconciliation. The eight-entry list totals214.07MW arithmetically versus the reported214.04MW; the regional total is not imported. Castella expansion remains held. CortijoI40.8MW and its1.2MW expansion are distinct excluded scopes; PNE42MW and14turbines are not assigned to either. None is presented as a current operating census.
Goldwind2025Q3 provides15explicit turbine columns, including table-reported165.8m and183.5m rotor diameters. Generator/converter voltages,50/60Hz choices, hub options, design life, operating/survival temperatures and environmental declaration claims retain their separate meanings. No unnamed component models or numerical curves are invented. SD3EX now retains three incompatible ratings3.2/3/2.5kW with a null scalar. Powersync25 gains exactPS25-240-1B identity, trip thresholds, enclosure and installation limits; conflicting inrush/fuse statements remain qualified. The viewer distinguishes offered hub options from scalar heights and unknowns.
The50-project and20-equipment objectives were not met:33projects and17equipment were substantively changed. Completed source partitions, field holds and original-byte hashes are in the research ledgers. No repeated source imports or synthetic variants pad those counts.
Polish project reconciliation and original environmental declarations#
Batch 2026-11-04-2435 adds six Polish projects and enriches the historical
Karcino record. All seven queued Polish references have now received detail-page
review. Legnica’s 2026 sale is separate from its planned 2030 commissioning.
Kuślin and Krzęcin II retain 2020 project-company, auction, permit and electrical
scopes alongside later launch observations. Proposed layout IDs are not installed
turbine records. Karcino’s 2009 historical year and 2010 developer launch remain
separate; the owner confirms Karcino and Karścino are different acquired assets.
All six additions and the enriched historical scope remain excluded from totals.
Original EPDITALY0508 and EPDITALY0852 declarations enrich seven existing Goldwind models. The recurring adapter pins original bytes, requires the complete explicit model set and preserves all earlier evidence and creation timestamps. Four swept areas and one design-life scalar are null pending revision reconciliation. Blade labels without explicit model allocation and unnamed components remain nested. No compatibility is inferred from a list’s order or similar model codes.
Ten modeled lifetime-generation scenarios retain their individual turbine, net-to-grid boundary, location and 20/25-year exposure. Egypt and Tanzania are exemplary LCA sites, not new operating projects. The viewer and offline example show these scenarios without summing them, deriving capacity factors or calling EPD verification independently validated turbine performance. No new curves.
All September 17–19 BOE A/B indexes were reviewed: zero wind notices. Hargraves’ primary developer page was acquired, but its three AEMO identities remain held without an explicit registry crosswalk. Next: September 14–16 BOE issues, the Wimmera stage boundary, Jasna’s exact installed fleet, and original Goldwind revision/type-certificate evidence. See the dated research ledgers for hashes, field locators and unresolved observations.
Dated EIA generator observations#
The complete final-2025 EIA wind Retired and Canceled tab contains 172 rows across 129 plant codes. It is not a complete historical retirement census. 58 existing U.S. project identities receive 82 dated generator observations; 71 identities remain deferred, including five code matches with name/location discrepancies. Whole-farm status, historical capacity, repowering boundaries and aggregation exclusions remain unchanged. The canonical extension retains row locators, original quantities, month/year dates, net seasonal versus nameplate MW, representative heights and predominant model labels. Those labels do not create exact OEM equipment or individual turbine records.
wind_catalogue.eia.parse_retired_partition verifies original workbook hashes,
reporting year, headers, wind prime movers, statuses and unique identities.
enrich_retired_partition requires an explicit reviewed crosswalk plus matching
WRI-encoded EIA code, normalized name and coordinates. Replaying a source
revision is idempotent; changing it under the same source ID is rejected.
The shared project viewer reports unit observations independently of farm
lifecycle and current capacity. The deterministic gallery prints their reporting
year and unchanged scope exclusions.
EIA’s workbook calls its speed field “Design Wind Speed”; its instructions ask
for representative annual average speed. Both descriptions are retained, with
measurement method, dates and height unknown. No measured/design performance
comparison or capacity factor is inferred. Sources: final-2025 EIA-860 and the
pinned WRI U.S. importer/identity helper (eia860_2025, wri_usa_identity).
The EIA adapter also accepts an inclusive operable_range=(low, high) plant-code
partition. The complete 1–60999 range is reviewed; 61000–61999 is next. Operating,
out-of-service and standby generator observations remain distinct in the shared
viewer and existing deterministic example. None establishes whole-farm current
operation, reconciles repowering, or promotes predominant model labels to exact
OEM equipment. The original final-2025 workbook bytes are reused and rechecked.
The 52000–55999 continuation reviewed all 61 plants / 67 generator rows in four
complete numerical partitions, including the empty 53000–53999 partition. It
adds dated evidence to 53 existing projects and holds eight unresolved identities.
The adapter rejects identical unit observations imported through overlapping
source aliases. Distinct tabs and revised observations remain separate. Viewer
cards separate reporting years; windEiaSummary(rows, year) can select one year
without treating absent observations as zero generating capacity.
The complete 56000–56999 partition adds evidence to 298 existing projects from 307 reviewed plants / 363 generator rows; nine identities remain held. Revised source observations can retain changed and unchanged units together without overwriting prior observations. Identical whole-plant partitions remain rejected when submitted under a second source alias.
The complete final-2025 Operable 57000–57999 partition reviews 266 plants and 290 generator rows: 258 existing project identities enriched and eight deferred. Reported generator observations do not change farm capacity or lifecycle totals.
The complete final-2025 Operable 58000–58999 partition reviews 143 plants and 173 generator rows: 132 existing projects enriched and 11 unresolved identities held. Original generator nameplates, seasonal net capacities, predominant model labels, representative hub heights and grid context retain their source rows and units. No whole-farm status, scalar capacity or aggregation eligibility changes.
A dated Nordex announcement adds the N175/6.X 7.3 MW mode and German tower
observations to the existing equipment identity (nordex_n175_update_20260422).
The 162.5 m hybrid tower was under development on 22 April 2026, with approval
scheduled for Q2; this is not evidence that approval or delivery occurred.
The same announcement reports Max Bögl’s 179 m tower approval on 21 April.
These options remain in extensions_json.oem_product_updates; prior scalar
specifications and provenance are preserved. The shared viewer counts dated
updates and models without this evidence. The offline example prints the modes,
tower conditions and dates without creating model variants or assuming installed
compatibility. No common-condition numerical curve or independently validated
performance is established by the announcement.
The complete final-2025 Operable 59000–59999 partition reviews 105 plants and 114 generator rows: 95 existing projects enriched and ten identities deferred. All prior scalar facts, source observations and capacity exclusions remain unchanged.
The complete final-2025 Operable 60000–60999 partition reviews 75 plants and 85 generator rows: 70 existing projects enriched and five identities held. The queued V82 archive power/Ct table remains deferred: its metadata labels it theoretical, provides no reference links and omits air density and operating mode. No conditions are reverse-engineered from Cp and no OEM curve is added.
The complete final-2025 Operable 61000–61999 partition reviews 85 plants and 100 generator rows: 67 existing projects enriched and 18 identities held. Jasna gains a dated mixed-fleet observation: 22 V126 turbines reported at 3.45 MW and 17 at 3.3 MW. The arithmetic check is separate from reported 132 MW, and no exact OEM revision or individual turbine identity is inferred. March 2021 commissioning-in-progress remains distinct from the prior launch label.
The complete final-2025 Operable 62000–63999 partitions review 146 plants and 161 generator rows: 36 existing projects enriched and 110 identities held. Karcino gains the operator’s dated history: turbine installation began in 2009 and commissioning occurred in 2010. The historical WRI conflict remains visible. The operator’s 120 GWh/year statement has no named observation year or method; it is not used to calculate capacity factor or presented as measured output.
Reviewed OEM history tables#
The wind_catalogue.history.stage_history adapter accepts a reviewed factual
transcription plus checksum-matched original bytes and a complete expected row
inventory. It preserves printed rating ranges, dated remarks and tower options
inside one type record. Explicit rating versions remain distinct; inferred
Cartesian variants are rejected by the review workflow. Existing models and
contradictory type/rating/unit rows remain in the decision ledger for separate
revision review. An OEM history table does not independently validate performance
or establish present-day availability. The November 5 Nordex transcription and
all 46 decisions are retained under catalogues/wind/research/.
EIA registry-site additions require independent identity review before staging. All generator observations remain nested with their original annual boundaries; missing farm capacity stays null. The shared viewer and offline example show unknown installed farm MW and unresolved aggregation scopes separately.
Dated OEM history and identity review (5 November, 09:10 batch)#
The 56 new US sites come from exactly 142 saved EIA identity holds, with 86 remaining deferred. Generic words no longer dominate name screening. Existing IDs, aliases, geodesic proximity and candidate-to-candidate proximity are reviewed separately. A clear screening result does not establish a physical farm boundary; all additions retain unknown farm lifecycle and capacities, with unresolved aggregation scope. The final-2025 Operable workbook scan is exhausted: 1,573 rows, maximum plant code 69,327, no residual code at or above 70,000.
Nordex’s dated 12 June 2025 retrospective supplies 31 exact-model enrichments. Prototype pitch and generator details apply to those prototypes. Discontinued production is a dated OEM claim and says nothing about installed-fleet retirement. The shared viewer counts these observations separately from current configuration fields, and the deterministic gallery prints their dates and boundaries.
The hash-pinned transcription adapter requires an explicit ID/manufacturer/model/ rating crosswalk and complete reviewed mapping. Identical replay preserves update timestamps; changing the same source observation is rejected. Scalar facts and creation timestamps are unchanged. The N175 narrative’s 200 m maximum and the undated table’s 199 m maximum remain separate observations. The N27 type/rating reversal and N169 printed-unit conflict remain deferred after reviewing independent dated OEM articles. A lone AW77/1300 narrative mention is held for original specification evidence. No new numerical curves or installed model variants are inferred from this history.
Published reference yields and saved EIA holds (5 November, 09:45 batch)#
Exactly 110 saved EIA holds from plant codes 62000–63999 were reviewed: 52 new registry sites and 58 remaining identity holds. The 36 previously enriched sites were not counted again. All new farm capacities and lifecycle states remain unknown and excluded from totals. Oregon PUC’s UM1401 report confirms one shared LGIA involving Orchard Windfarm I, 2, 3 and 4; it does not supply the EIA crosswalk or resolve their physical boundaries, so those four candidates remain held.
FGW’s current index supplies complete Nordex tables for the old and new reference sites: 110 and 70 rows respectively. They add 16 explicitly printed types/revisions and enrich 24 existing equipment records. Hubs and power settings stay nested, including 173 accepted row observations; seven N27 rows remain in the conflict ledger. Five withdrawn N149 yields are retained alongside replacements with the publisher’s withdrawal and replacement dates. Three missing rotor cells stay null. Existing scalar specifications are unchanged when reference-table geometry differs.
The offline wind_catalogue.reference_yield.stage_reference_yields adapter checks
original-byte hashes, complete row inventories, explicit record-ID crosswalks,
positive finite quantities and restriction notes. Identical replay preserves
creation and update timestamps; changed observations under the same source ID
are rejected. Original decimal-comma/thousands-dot strings remain reviewable.
FGW defines these quantities as calculated five-year energy in kWh. They describe hypothetical reference sites, not measured plant generation. Old-site logarithmic and new-site Hellmann profiles are kept separate; density, operating mode and gross/net boundary remain unknown. No annual generation, capacity factor or numerical power/Ct curve is inferred. The shared viewer and deterministic gallery show counts by reference site, withdrawn observations and missing evidence without summing yields or presenting an unsupported performance comparison. Paid reference certificates are separate from the freely readable published tables.
Korean certificate configurations and Salinas III proposal (5 November, 14:35)#
The remaining 55 rows of the checksum-pinned KNREC register are now reviewed: 23 certificate-scoped equipment configurations are added and 32 rows await revision crosswalks or conflict resolution. Together with the prior two HTq rows, all 57 registry rows have a recorded disposition. Multiple power and hub values remain options within one record, never Cartesian variants. Printed expired notes are retained; blank status does not establish current certification validity. The shared viewer and offline gallery distinguish these registry observations from independent performance validation. Six new OEM labels retain original Korean company names without inferring corporate succession.
BOE daily indexes for 14–16 September 2026, including both Series A and B and any linked extra issues, yield one new wind-only proposal: Hibridación FV-PEol Salinas III (PEol-FV-313). Its reported 36 MW planned capacity, eight explicitly identified proposed turbines, 118 m hubs, 163 m rotors, 30 kV underground collection and EUR 39,576,856.27 material-execution budget are source-qualified. The budget price year is unknown. Existing PV capacity and the proposed hybrid total are separate observations. Public consultation establishes neither approval nor installation. The eight annex turbine IDs link to the proposed parent, with unknown coordinates. The named base V163 model is recorded as a relationship; an exact equipment-configuration link remains unset because the planned 118 m hub differs from the existing OEM page’s published hub options. Private landholder names are not copied. Taraguilla’s paragraph-number correction was already retained and is counted as duplicate evidence, not an enrichment.
REPD offsets 600–800 received structural review only: all 200 applications remain
identity holds, with no claim that original permit contents have been reviewed.
No capacity aggregation exclusion is lifted. All earlier records and creation
timestamps remain unchanged. See the dated KNREC, BOE and REPD decision ledgers
in catalogues/wind/research; the continuation cursors are in backlog.json.
Brazil: dated ONS unit evidence#
The first 100 lexically sorted wind CEGs in the ONS daily file retrieved on
2026-11-05 contain 66 exact historical WRI identifier/name matches and 34 identity
holds. The WRI Brazil importer explicitly encodes the six-digit numeric CEG
component as BRA plus seven digits. The adapter additionally requires one
historical identity, exact name agreement after case/whitespace normalization,
and no competing CEG version anywhere in the file. Missing identities stay held.
The source dictionary calls val_potenciaefetiva nominal generating-unit MW.
Unit rows may represent turbine groups. Nested observations retain the original
values, owner/operator, state, subsystem, operation modality, release-for-test,
release-for-commercial-operation and deactivation dates. Missing dates remain
null; retrieval and metadata modification dates are not observation dates.
No unit count becomes a turbine count, release date becomes farm commissioning,
or unit MW becomes whole-farm installed capacity. All 66 historical aggregation
and lifecycle exclusions remain. The shared viewer and offline gallery count
unit evidence and distinguish missing nominal ratings from reported zeros.
This is an ONS-dispatched-unit partition (modalities I/II-A/II-B/II-C), not a
current operating census of Brazilian wind farms. Original CSV/dictionary hashes,
per-row locators, candidate decisions and the next exact CEGs are in the source
registry and research/2026-11-05-1510-ons-review.json. The daily source has no
historical archive guarantee; continuation must use the pinned bytes or explicitly
record a replacement snapshot and reassess partition boundaries.
The next five complete 100-CEG partitions (offsets100–599 of the same pinned snapshot) add 99 existing-project enrichments and retain 401 identity holds. The first600CEGs are now reviewed; no source release or lifecycle is silently advanced. Adjacent-partition tests ensure complete multi-unit CEG groups are neither split nor duplicated. The gallery also reports source-state coverage as snapshot counts, never as installed capacity.
The final 482 CEGs (offsets 600–1081) add 175 project enrichments and retain 307 identity holds. The pinned scan is now complete: 1,082 CEGs reviewed, 340 matched historical projects and 742 unresolved identities. Completion describes this source partition only. The resume auditor rejects gaps, overlaps, incomplete terminal groups and changes of source hash or retrieval version. The viewer and gallery count each project once per reported state across its observations; unknown geography remains visible and no current fleet distribution is implied.
Goldwind Americas 2.5 MW tables#
The complete English three-model and Spanish two-model tables support three explicit models: GW 100/2500, GW 109/2500 and GW 121/2500. Shared table cells are applied only to their printed columns. Spanish edition details remain nested, with their own provenance and units, because option/revision equivalence is not established. Hub choices are not expanded into additional models.
Original reported swept areas and diameter-based checks are retained separately; uncertain area scalars remain null. GW 109/2500 also retains the English static 10.5 m/s and Spanish unqualified 10.2 m/s observations with a null canonical rated wind speed. Neither brochure resolves the Korean certificate’s 107.5 m rotor crosswalk. The five plotted power curves are graph-only, with density stated but mode unresolved; no numerical curve, compatibility claim or independently validated performance is inferred. See the 1735 OEM decision ledger and source registry for checksums, exact page/column locators and continuation tasks.
Dated EPE-hosted SIGA and Spanish proposal observations#
The 2026-11-05 continuation reviews the complete 381-entry Bahia wind
partition of EPE’s hosted SIGA layer, whose dataset date is 2026-01-13.
Repeated object-ID queries agreed before and after all pages were acquired.
Sixty-eight entries enrich existing projects through a unique WRI-encoded CEG
identity plus exact name; 313 identities remain deferred. The original CEG
spelling is retained, with only numeric suffix padding compared. The layer’s
misleading date-typed idenucleoceg field is not used as an identifier.
Authorized kW, inspected kW and physical-guarantee kW are normalized separately to MW inside dated observations. Physical guarantee is not installed capacity. Original lifecycle text, operating-entry date, permit term, ownership text, municipality and registry-point coordinates remain qualified by their source. Coordinate decimal precision is not positional accuracy. None of these observations changes canonical capacity, lifecycle or aggregation eligibility. The shared viewer and deterministic gallery show category completeness, including reported zeros and unknown values, without summing incompatible MW. This successful public EPE ArcGIS download is independent of the earlier ANEEL CSV HTTP 503 failure; neither failure implies authentication or a reuse ban.
BOE Series A/B issues of 10–12 September 2026 yield two proposed Salinas wind modules, each 36 MW with eight individually identified planned turbines, and the archived 144 MW Alto Cabrera application. Proposed wind MW, existing solar MW and combined hybrid MW remain separate. Alto Cabrera’s whole-project status is unknown: dismissal of application PEol-382 does not establish the fate of all possible iterations or appeals. Salinas electrical length/conductor inconsistencies and the unresolved 118 m OEM hub configuration are retained. The solar-only environmental notice remains an identity hold, and the generic cadastral notice is excluded. Private landholder names are not imported.
Four exact Goldwind 1.5 MW brochure models are added, and two existing 2.5 MW models receive separate digital-English edition observations. The latter explicitly distinguishes survival from operating temperature. Printed rotor areas and earlier rated-speed disagreements remain qualified observations; unresolved scalars stay null. Six graph-only curves are deferred, with no numerical points or synthetic variants created. These are manufacturer specifications, not independently validated performance.
Sources: EPE SIGA layer, Goldwind 1.5 MW original, Goldwind 2.5 MW digital original, Alto Cabrera notice, Salinas II notice and Salinas I notice. Full bibliographic metadata and stable source identifiers are in the reference catalogue.
Rio Grande do Norte continuation and bounded OEM review#
The complete 317-entry Rio Grande do Norte EPE wind partition is now reviewed. The before/after ID inventories agree, and both response pages are complete. 107 existing projects gain dated observations; 210 identities remain held (200 missing or nonunique historical matches and 10 ID/name disagreements). The dataset date remains 2026-01-13, distinct from retrieval on 2026-11-05. Authorized capacity, inspected capacity and physical guarantee remain separate observations. All six eligible capacity summaries are unchanged. Suspicious source operation dates, including Macau’s 1900 date, stay in the original observations and are not promoted to commissioning milestones.
The adapter now rejects a CEG revision change against prior EPE evidence even when the project has no scalar ANEEL identifier. A real two-row RN fixture tests separate capacity categories, reported zero and the suspicious date boundary. The shared viewer and deterministic gallery show distinct linked projects by state and dataset date, with unresolved identities and aggregation exclusions visible. This is selected evidence coverage, not a current operating census.
Three exact Windey model records are added: WD200-7700 (Saudi 60 Hz), WD220-10000, and WD225-9000. Their explicitly reported unit ratings are 7.7 MW, 10 MW and 9,000 kW respectively. Detailed configurations are partial; rotor diameters are not inferred from model names. Twelve OEM case cards and two further announcements remain queued for project, revision or certificate crosswalks. The Saudi configuration is not assigned to the Ethiopian proposal. No numerical curves or independently validated performance are added.
Bachmann’s component tables and case study are excluded under its explicit prior-consent reuse terms. Delta’s proposed PDF redirects to a terms-verification page with an internal-use restriction; the original PDF was not acquired. Ingeteam’s two historical document links redirect to homepage HTML. These are recorded separately from authentication or HTTP download failures. Windey’s large-image timeouts were resolved, with all four original image checksums saved in the review ledger. The equipment target was not met by inventing variants.
This run adds 0 projects, 3 equipment records, 0 individual turbines and 0 curves, and enriches 107 projects. Branch totals are 8,294 projects, 489 equipment records, 13,657 turbines and 2 curves across 59 countries and 64 equipment-manufacturer labels. The next registry batch is the complete EPE CE partition; previously completed BA/RN and ONS partitions stay closed.
Sources: EPE SIGA layer,
Windey product cases,
Windey Saudi announcement.
Decisions and original-byte checksums are in
2026-11-05-1935-epe-review.json and 2026-11-05-1935-oem-review.json.
The November 5 20:35 continuation reviews all 103 Ceará EPE wind entries, linking 53 existing identities and holding 50 for a historical/phase crosswalk. The dated categories do not change capacity eligibility. BA and RN partitions are already complete; the next complete EPE partition is RS.
The Windey June 2022 English brochure contributes 31 exact models and one
historical observation for the existing WD225-9000. PDF pages 7, 8 and 12
(printed 09/10, 11/12 and 19/20) contain 32 explicit columns; pages 9–11 remain
queued. The checksum-pinned oem_tables adapter consumes visually reviewed
columns and rejects missing, reordered or duplicate identities. Generator RPM
and converter kVA stay separate from rotor RPM and turbine kW. Temperature,
cut-out and hub-height ranges remain options; the viewer and gallery expose
these without claiming installed configurations or validated performance.
Complete Rio Grande do Sul partition and remaining Windey columns#
The November 5 21:35 continuation reviews all 80 EPE-hosted SIGA wind entries for Rio Grande do Sul. Stable before/after OBJECTID queries and the complete response agree. 67 existing projects gain dated native registry observations; 13 identities remain held. No projects, individual turbines, capacity eligibility or lifecycle claims are inferred. All six capacity summaries, including unknown and excluded counts, remain unchanged.
The remaining Windey brochure tables on PDF pages 9, 10 and 11 (printed 13/14, 15/16 and 17/18) add 35 exact models. All 67 printed model columns on PDF pages 7–12 have now been reviewed across two batches. The prior source registry and original-byte checksum remain intact; the new review ledger records expanded locators and visual transcription. These are historical specifications, not independent performance validation or installed project configurations.
Column boundaries matter: WD164-3600 states a converter range of 47.5–51.5 Hz, while the other 34 newly reviewed columns state 47.5–52.5 Hz. WD172-5500’s 80.5 m blade length is retained as printed, without replacing it using another 172 m rotor model. Generator kW/RPM, converter grid-side/turbine-side kVA and A, component mass/dimensions, hub-height options and operating conditions remain separate observations. Exact component model IDs, density, operating mode, warranty, certification identifiers and numerical curves remain unknown.
The offline adapter now rejects malformed component quantities and vectors. Small attributed fixtures exercise column-specific values and unit boundaries. The shared viewer counts usable converter rating pairs separately from missing ones, rejects malformed hub ranges, and explicitly separates component quantities from turbine ratings. The gallery dynamically displays the reviewed model count. The next complete EPE partition is PI; the next OEM/project work is the AfDB Aysha I original ESIA and exact phase crosswalk recorded in the backlog.
Piauí registry continuation#
The November 5 22:35 batch reviews the complete 126-entry EPE Piauí wind partition, with stable before/after object IDs and original response checksums. Thirty-six existing identities receive dated registry observations; ninety remain held. No current operating status, new farm or capacity eligibility is inferred. The existing state-coverage gallery and viewer now include Piauí.
Replay the reviewed ledger without changing the catalogue:
uv run python scripts/replay_wind_epe.py catalogues/wind/research/2026-11-05-2235-epe-review.json
The command rejects changed observations or changed identity decisions. Next is the complete Pernambuco (PE) partition. Independent Aysha I follow-up located IFC project 49251 and obtained the operator-hosted ESIA original; detailed phase and equipment reconciliation remains pending. No equipment is added from model names or marketing claims, and no completed Windey table is rescanned.
Pernambuco, Paraíba and Santa Catarina continuation#
The November 5 23:35 iteration reviews three complete EPE-hosted SIGA wind partitions dated January 13, 2026: Pernambuco 46 entries (31 linked, 15 held), Paraíba 42 (12 linked, 30 held), and Santa Catarina 18 (10 linked, eight held). It enriches 53 existing projects with separately qualified registry observations. No project, turbine, equipment or curve is added. All capacity summaries and historical baseline exclusions remain unchanged. The 50-project objective is met; the 20-equipment objective is not met.
The shared viewer now counts unresolved scope and unknown lifecycle among projects with SIGA evidence. The deterministic gallery compares linked versus held identities for complete selected state partitions. Neither chart represents a current national operating inventory. Completed states are not rescanned.
Original response verification is optional and read-only:
uv run python scripts/replay_wind_epe.py catalogues/wind/research/2026-11-05-2335-epe-review.json --originals /path/to/original-responses
This verifies SHA-256 for every retained response, reparses both ID queries and all pages, and compares the complete candidate transcription before replaying accepted records and decisions. Missing files, changed bytes, drifting IDs and changed transcriptions fail. Original bytes stay outside Git; URLs, hashes and review decisions are committed.
Independent Aysha I research found unresolved dated capacity observations: IFC’s August ESRS reports 300 MW; its October investment disclosure reports 300.3 MW; MIGA’s portfolio identifies 306 MW. MIGA portfolio Phase II is not the Aysha II wind farm. IFC describes a May ESIA but lists a June attachment; that original attachment remains unavailable for revision comparison. These are research holds, not canonical capacity replacements or exact equipment links. IFC HTML HTTP 403 and missing attachment links are access limitations, not assertions of authentication or reuse restrictions. Next: Maranhão’s complete EPE partition and the exact IFC June ESIA attachment/phase crosswalk.
Completing the captured EPE roster#
The November 6 00:35 batch reviews MA (16 entries) and MG, PR, RJ, SE and SP (one each). Two existing projects gain dated registry evidence (RJ and SE); 19 identities remain held. Thirteen remaining states return explicit empty ID lists. A separate null/unknown-state query is empty. Two national ID queries agree on 1,134 entries, exactly matching the disjoint union of all 27 reviewed state partitions. Across this captured layer, 386 entries are linked and 748 are held. This establishes source membership coverage, not a current Brazilian operating census or completeness of all Brazilian wind assets. No capacity eligibility changes; the 50-project and 20-equipment objectives are not met.
uv run python scripts/replay_wind_epe_coverage.py catalogues/wind/research/2026-11-06-0035-epe-coverage.json --originals /path/to/original-responses
The coverage check verifies reviewed candidate IDs, original query checksums, explicit empty states, disjoint partitions and matching national rosters. Missing/null/error/truncated responses are unknown, never zero. The offline fixture preserves an actual empty AC response. The gallery lists these empty partitions separately; the viewer distinguishes valid source registry points from absent coordinates, without borrowing historical WRI points or treating coordinate precision as positional accuracy.
BOE September 7–9, all discovered Series A/B issues, yields two held notices: Viento de Sellía parcel correction and Colladillo/Cabeza de Mesá solar hybridization. Neither supplies an accepted wind identity crosswalk or complete turbine inventory. Original notice and index checksums are retained. Exact IFC June ESIA and WD230-11500 certificate searches did not locate the requested originals; prior announcements are not counted again. Next: BOE September 4–6 and the saved MA Delta 3 identity/phase crosswalks. Completed EPE partitions stay closed.
September 4–5 BOE modification and repowering decisions#
The November 6 01:35 continuation adds Alcocero de Mola’s revised 78 MW proposal and the separately authorized 5 MW Boyal I and Boyal II replacement generations. All three remain excluded from capacity totals. Environmental clearance alone does not establish construction permission; repowering permission does not establish replacement operation. Boyal’s historical 4.5 MW units, proposed 5 MW units, 4.5 MW grid access and reported 15 MW transformer active-power limit remain distinct. No anonymous turbine children or exact installed OEM configurations are inferred.
The original Alcocero notice reports 197.92 m tip height versus 197.5 m from hub plus radius, and 1,371 m of line versus 1,370 m of listed segments. Both observations are preserved. Boyal II repeats a Boyal I AESA reference; no permit identity is silently corrected. September 6 returned HTTP 404 and is not a completed empty issue. Next daily partition: September 1–3. The saved Delta 3 crosswalk remains held after the original ANEEL notice returned HTTP 403; this is not evidence of authentication or a reuse restriction.
Complete daily issue replay#
review_daily_indexes verifies all discovered ordinary and extraordinary issue
URLs before returning Series A/B wind-title candidates. Missing issues fail even
when the default page has no candidates; identical notice aliases are deduplicated
and conflicting titles are rejected. This is title-scan completeness, not a wind
project census. The September 1–3, 2026 scan includes both September 2 issues and
returns no candidates. See 2026-11-06-0235-review.json for source hashes and the
separate, unresolved Wimmera survey/permit phase review.
Original refusal history and permit envelopes#
The August 27–29 and 31, 2026 BOE batch enriches Satrebares from its original refusal notice (BOE-A-2026-18277), preserving the later geography correction. Application/hearing dates, the received grid-access expiry notice and electrical proposal observations do not describe installed assets. The adapter rejects operational, permitted or retired status inferred from an application refusal. The Valdepero solar and Frontones battery notices are not new wind capacity. Wimmera’s acquired PA2000877-2 permit provides upper/lower dimensional limits, not exact turbine configurations or AEMO phase allocations. Its amendment date is May 14, 2026; May 19 is the attachment filename date. See the 0335 review ledger.
Floating prototype and OEM numeric tables (6 November, 09:35 UTC)#
The complete discovered BOE July 30–31/August 1 A/B partition yields WHEEL’s preliminary authorization and San Jorge’s refused wind application. Neither is installed capacity. Two solar decisions remain held for wind identity crosswalks. WHEEL is linked to its exact PLOCAN pilot substructure configuration, supported by the original 2025 environmental decision and 2026 authorization. Floater seabed elevation (−90 m, Port of Las Palmas datum), approximate 100 m anchor depth and 70 m minimum selection criterion remain different quantities. Construction authorization is expressly conditional; port fabrication is not proof of installed offshore operation. The shared PLOCAN 15 MW grid access is not added to the prototype’s 6 MW design. ELICAN is a shared terminal, not an inferred parent or phase. No exact turbine model was supplied in these notices.
All eleven exact types in SANY’s global English navigation were reviewed. Ten models are added; SI-16840 is held against the existing India configuration pending a revision and legal-entity crosswalk. The numerical table adapter requires a complete explicit model list, units, source and locator, and preserves missing cells. Three inconsistent cells remain null: SI-16850 cut-out speed, and SI-18572/SI-19580 swept area. Circular-area arithmetic is a labelled check, never a replacement specification. Generator voltage is not grid voltage; operating and survival temperatures are separate. IEC claims are not verified certificates; specifications are not independently validated performance.
The OEM’s original endpoints returned HTTP 508 with empty bodies. Public indexed OEM tables supplied the evidence; original-byte checksums, publication dates and revisions remain unavailable. No authentication or reuse restriction is inferred from the failure. Unnamed blade and generator platform ranges are held. The independent South African IPP inventory returned HTTP 404 after a login redirect; no login was attempted and no inventory was imported. REPD saved candidates 1000:1200 were structurally reviewed: all 200 remain held for original permits.
The shared viewer exposes OEM table completeness and withheld fields. The offline gallery distinguishes floating design depth boundaries and prints the retained OEM conflicts. Next: BOE July 27–29, REPD 1200:1400, and exact SANY corrections.
6 November continuation: empty discovery partitions and restricted originals#
The July 27–29 BOE Series A/B title partitions are complete with no wind-title candidates. This does not establish that no wind projects existed on those dates. REPD saved candidates 1200–1399 were structurally reviewed against the pinned release; all 200 remain application-to-farm identity holds.
The original SANY Product Family, April 2025 brochure was acquired and
checksummed, but its explicit copying restriction excludes its specifications
from this intake. This is an explicit reuse exclusion, not an HTTP or
authentication failure. The three existing SANY scalar conflicts and the
SI-16840 regional configuration crosswalk remain unresolved. No projects,
equipment, turbine instances or numerical curves were added in this continuation.
See catalogues/wind/research/2026-11-06-1035-review.json for source identifiers,
checksums and next partitions.
Operator portfolio and BOE continuation (6 November, 11:35 UTC)#
The complete 20-row operational wind subset in Masdar’s 15 July 2024 offering circular was reviewed against the catalogue: nine new projects, three historical observation enrichments, and eight identity holds. The portfolio date is 31 March 2024. It is not a current operating census. Contracted MW (pp154–155) and gross reported MW (p166) remain separate nested quantities; neither becomes installed capacity. Turbine count × rating is a labeled arithmetic check. The source’s fleet-average footnote is preserved and does not identify an exact model. No equipment configurations were inferred from OEM names or ratings.
Dumat’s 2022-07-01 commercial operation date and March 2024 ownership are retained alongside its original 2019 award observations. London Array’s final-turbine installation and Hywind’s first-export milestones are not relabeled full COD. Month-only African CODs retain month precision. Existing scalar facts, IDs, creation dates and capacity exclusions are unchanged. The shared viewer shows dated portfolio coverage, missing exact models and capacity boundaries.
BOE July 23–25 contains one accepted wind application: Revilla-Vallejera Hib, 40.5 MW proposed, linked by procedure PFot-ALM-044-EOL. Its operational PV/BESS host and shared grid capacity are excluded. Two source reference coordinates remain in ETRS89 UTM zone 30N with metre precision; neither is a farm centroid or an individual turbine inventory. FV Monclues is a solar-only authorization and remains excluded from new wind projects. REPD offsets1400:1600 received structural review only: all200 identities still need original permits.
Continuation: BOE July20–22; REPD1600:1800; eight named Masdar registry crosswalks in the review ledger. The same circular’s Zarafshan/Baltic Eagle construction rows are a separate unprocessed partition. Full original documents are not redistributed; SHA-256 and page locators are recorded in the source registry and review ledger.
Construction and environmental decisions (6 November,12:35 UTC)#
Masdar’s two construction-table wind rows from its July2024 circular are now represented: Zarafshan and Baltic Eagle. They describe March2024 construction, not present operation. Zarafshan’s2025 table forecast conflicts with its2026 narrative forecast; both remain dated observations and no COD scalar is assigned. Baltic Eagle’s minority-stake financing is not total project investment.
BOE July20–22 yields one negative environmental decision for two wind proposals, Escuderos I andIV. Their21.6/28.8MW and V172 geometry are original proposed designs. The negative DIA is neither construction permission nor an inferred final administrative cancellation. PV hosts and shared grid infrastructure are excluded. All four records remain excluded from additive capacity totals. No equipment variant is manufactured from project ratings or optional hub configurations.
The operator adapter now requires a reported lifecycle before staging a row. A regression fixture checks that construction and conflicting forecasts cannot become operation or commissioning. REPD1600:1800 was structurally reviewed only; all200 original-permit identity holds remain. Next: BOE July16–18, REPD1800:2000, and later Zarafshan/Baltic Eagle milestone and rating-revision reconciliation.
July 16–18 notice and REPD continuation#
The 6 November 2026 13:35 review completed all discovered BOE issues for 16–18 July. Notices BOE-B-2026-24346 and BOE-B-2026-24461 concern solar hybrid applications; La Herrada and Torozos B/C remain wind identity holds. Solar, wind and hybrid capacities are not interchangeable. REPD saved candidate offsets 1800:2000 were structurally reviewed, with all 200 held pending original permits. The offline parser now rejects extra or missing CSV cells before recording a reviewed partition, including malformed rows outside the selected slice. No project, equipment or capacity total changed. The dated review ledger records the original Baltic Eagle installation release and its unresolved date/rating revision crosswalk.
The second evening checkpoint (2026-11-07-2040) completes BOE June 29–30:
Benavieja (PEol-947) is a refused 86.8 MW proposal, not an operating plant;
Iglesias storage and San Lorenzo C solar notices remain wind-identity holds.
Three CRRC Times New Material tables contribute 19 explicitly printed blade
configurations (18 model names). The two 3.XMW-76 columns retain their distinct
wind classes and root pitch-circle diameters. Root geometry is not rotor geometry;
MW prefixes do not establish a component electrical rating. Certification text
remains an OEM claim with unknown certificate identity and validity. No turbine
compatibility, material variant or current certification is inferred. Original-byte
acquisition returned HTTP 403; official search-rendered tables were reviewed, and
checksums are explicitly unavailable. The linked legal page renders as a missing
column; no authentication requirement or explicit reuse prohibition was established.
Swedish offshore application areas#
The offshore adapter requires every layer 41–50 plus independently fetched counts.
It retains the complete polygons, application dockets, organization roles, dates,
planned/approved/erected turbine counts and planned/calculated generation separately.
InstallEff also occurs on refused proposals; it remains an unclassified MW
observation, never an installed-capacity substitute. Erection does not establish
current operation, and an inactive application does not establish retirement.
No individual turbines or fixed/floating foundation types are inferred.
Repeated names/IDs and prior aliases or locations inside a conservatively buffered bounding box are held for reconciliation. Bounding-box intersections among new areas are review candidates, not established phase or repowering relationships. The shared viewer and deterministic gallery show application stages, unknowns, reported zeros and aggregation exclusions without summing these MW observations.
ABB wind-converter electrical tables (2026-11-08)#
The complete converter/module and board partitions in ABB’s ACS800-67 hardware manual (3AFE68392454 Rev C, effective 2010-05-11, pp. 89–91) and ACS800-67LC manual (3AUA0000058400 Rev C, effective 2018-11-09, pp. 121–123) add 20 exact converter types and two control boards. Original PDFs were acquired and hashed; only attributed factual transcriptions are committed. These are dated OEM specifications, not independently validated performance or current availability.
Whole converter assemblies, aggregate grid/rotor converter types and physical modules remain distinct. Only explicit table relations link their stable IDs. The ACS800-104-0580-7 grid-side DC power and rotor-side AC power are separate observations. Apparent power in kVA is never silently converted to active kW. Duty cycle, temperature, frequency, voltage and derating conditions stay beside the ratings. A dash remains null, including the APBU-44 maximum current. No exact wind-turbine compatibility or numerical power/Ct curve is inferred.
electrical_tables.stage_electrical_tables verifies original checksums, complete
reviewed model partitions, circuit roles, units and explicit application links.
Existing model identities require revision review. The synthetic offline fixture
exercises missing rows, duplicate models, unknown versus zero, invalid quantities
and dangling compatibility. Wildcard filters and equipment without a suitable
category (fuses/fans) remain deferred rather than classified incorrectly.
References: abb_acs800_67_revc_2010 and abb_acs800_67lc_revc_2018
in the generated reference catalogue.
The 2026-11-08 NVE partition reviews all 72 wind-plant API identities against
its 65-entry operating subset. It adds 49 projects and holds 23 possible overlaps
for explicit registry crosswalks. TurbinID identifies returned cohorts: a row
with multiple turbines is never expanded into invented unit identities. Plant
mean geometry and first-stage commissioning dates retain their original scope.
Normal-year GWh are modeled observations with unknown climate period and gross/net
boundary; they do not supply a capacity factor. All new scopes remain excluded
from capacity totals. The shared viewer exposes missing coordinates, mixed model
labels, unknown versus zero output, and cohort-count disagreements.
ABB ACS800-67LC Rev C pages 124–131 add three exact lcl_filter types and enrich
five existing converter assemblies. Cooling, heat loss and dimensions retain
option-specific conditions; unweighted noise is not promoted to dBA sound power.
The dated manufacturer’s declaration is not independent certification. Original
ratings, stable IDs, creation dates and prior evidence remain intact.
The June 8–13 BOE partition separates Mimas, Titán and Dione wind modules from solar assets and shared evacuation. El Casarejo remains a proposal; Buenavista’s dated technical inventory does not establish current operation. All five scopes remain excluded from capacity totals. The ACS800-67 Rev C enrichment preserves alternative warranty clocks (installation versus manufacture), indoor conditions, and old/new module incompatibility; no scalar warranty or new model is inferred.
Auxiliary cooling and permit boundaries#
The November 9 evening batch retains the W2E143-AA09-01 cooling fan’s 50 Hz and 60 Hz electrical-input observations separately. Fan watts are not turbine rated kW, and fan geometry does not enter turbine rotor/specific-power summaries. The 2022 OEM revision and the 2018 ABB type mention do not prove an installed revision or interchangeable replacement. Assembly type references are evidence observations; exact revision compatibility remains unknown.
Campanario II híbrido’s construction authorization is not commissioning. Mazo’s withdrawal applies to the reviewed application, not retirement of an operating farm. Both records remain excluded from capacity totals.
Exact connector counterparts (10 November 2026)#
The complete numbered ABB ACS800-67LC Rev C counterpart table (pp126–127) contains 41 terminal/part rows and 22 distinct HARTING part numbers. Each exact part now has its current public OEM product table, original HTML checksum, section/attribute locators and dated links to the five existing ABB assemblies. Prior assembly observations and creation timestamps are retained. Unnumbered cable glands remain excluded. No matching cross-sector equipment IDs were found.
The connector category includes inserts, contacts, connector assemblies and
hoods. Ratings retain original units, inequalities, conductor/earth versus
conductor/conductor voltage, UL/CSA conditions, temperature limits and conditional
IP protection. Superscripts are represented as ^(...), so resistance greater
than ten to the tenth is not flattened into 1010 ohms. OEM claims are not
independently validated performance. The product name and size labels for
19340060421 differ (Han6B versus size16B); both are retained as published.
Current OEM revision and historical installed revision remain unknown; exact part-number crosswalks establish dated references, not verified installation or compatibility. Electrical derating arrays are held for separate axis/condition review. The shared equipment viewer and deterministic example report counterpart coverage, unknown quantities, reported zeros and unverified revisions without summing electrical ratings as turbine output. The offline connector adapter rejects homepage redirects, mismatched parts, duplicate attributes and incomplete numbered partitions. Its tests use an explicitly synthetic HTML fixture.
The BOE May18–23 Series A/B title partition adds PE Orzar and PE Anoia i Segarra III. Both retain planned capacity and unresolved aggregation boundaries. Orzar’s 56-to-52 MW authorization history and partial hub-height modifications remain dated observations. Three solar-related corrections are held for original wind identity reconciliation; they do not change wind lifecycle. Cadastral personal data are omitted.
Proposed layouts and repowering reconciliation (10 November, evening)#
This batch adds five projects, enriches the existing Fakken project, adds five individually labelled proposed turbine positions and enriches one existing HARTING part. It does not meet the 50-project or 20-equipment targets. No configurations were inferred to fill those targets. All five new projects remain excluded from capacity sums; Fakken retains its existing exclusion.
The complete discovered BOE May 11–16 Series A/B wind-title partition contains three notices. Pocitos records authorization history for 20.79 MW. Its Arico line burial is auxiliary infrastructure, not another wind project. Armus is a public consultation for a 35 MW wind module, separate from existing solar capacity and shared evacuation. Five explicit AR_01–AR_05 positions retain ETRS89/UTM zone 30 coordinates in metres. The met mast is excluded. Geographic coordinates remain null until a reviewed transformation; the viewer shows these unmapped positions and their proposed status. Nordex is named, but the notice supplies no exact model; 175 m rotor diameter does not establish an N175 configuration.
The NVE 2018 letter, NVE history, case 4790 and owner history connect Fjeldskår’s original five turbines to the two-turbine Lindesnes replacement. Saved NVE registry IDs 1 and 66 retain their distinct generation identities and a replacement link. The original 3.75 MW, proposed ceiling of 8 MW and operating replacement’s 7.2 MW remain separate. July versus August 1998 commissioning and 2018 cessation/demolition versus the registry’s 2019 end date are retained as dated conflicts; exact old commissioning and retirement dates are null. Cohort IDs are not treated as individual turbine inventories.
Fakken’s operator and NVE plant 9333 corroborate the current 54 MW/18-turbine phase. Its historical WRI 59 MW and stable identity remain intact. Fakken II is an additional proposed 66 MW/11-turbine phase, not repowering. The operator’s combined annual production claim is not assigned to that phase or converted into a capacity factor. Exact fleet revisions and period-qualified production remain unresolved.
The HARTING drawing for 19340060421 is OEM-authored and hosted by a distributor. Rev B dated 17 January 2017 supplies labelled millimetre dimensions and an M25×1.5 thread. These are visual transcriptions of labels, not digitized geometry or independently validated performance. The “6B” title versus “16 B” current product-table label and historical installed revision remain unresolved. No compatibility is inferred.
Original bytes were acquired and checksummed outside Git. The batch review ledger records notices, additions, enrichments, conflicts and exclusions. The offline layout adapter rejects mismatched checksums, unreviewed CRS, partial or duplicate position lists, malformed coordinates and operational parents. Continuation: BOE May 4–9; NVE Bessakerfjellet registry20, then Havøygavlen3; HARTING’s historical installed revisions for the saved 22-part partition. Do not rescan the completed NVE inventory or HARTING product tables.
2026-11-11: Norwegian identities, Anjana layout and Nexans cable configurations#
This iteration adds Anjana and Skomakerfjellet, enriches the existing WRI Bessakerfjellet and Havoygavlen identities, and adds 24 exact Nexans catalogue configurations and 17 proposed Anjana turbine positions. It does not meet the 50-project target. The canonical totals are 8,475 projects, 657 equipment records, 13,679 turbine records and two original numerical curves in 64 countries, with 70 equipment manufacturers. These are catalogue records, not an operating census.
The complete BOE May 4–9 Series A/B wind-title partition contains four notices. BOE-A-2026-9683, BOE-A-2026-9684 and BOE-B-2026-14190 concern solar/battery hybridisation and are excluded as new wind projects. BOE-B-2026-14921 establishes Anjana’s proposed 86.8 MW, nine 5.2 MW modes plus eight 5 MW modes, and a published 17-position ETRS89/UTM30N layout. The separate reserve position and two met masts are excluded. East/west counts conflict within the notice; no normalized phase assignment or exact turbine OEM is inferred. Original projected coordinates and centimetre publication resolution remain distinct from positional accuracy.
NVE cases 22 and 259 and Aneo’s separate project pages distinguish Bessakerfjellet from Skomakerfjellet (formerly Bessakerfjellet II). The former retains its WRI ID, creation date, 58 MW historical rating and original point. Its current 57.5 MW and Skomakerfjellet’s 13.2 MW are independent farm capacities; the 14 MW Skomakerfjellet permit envelope is separate. First-stage cohort dates are not whole-farm COD.
Finnmark Kraft reports 41.4 MW at Havøygavlen, compared with 40.8 MW in the saved NVE registry. Its stated nine 4.3 MW units plus a 3 MW Siemens unit yield a labeled 41.7 MW arithmetic check. No value replaces the reported figures. Installed MW is null, the record remains excluded, and the retired Nordex and replacement cohorts are retained without additive aggregation or a derived capacity factor.
Nexans’ complete TSLF-O card partitions 4320, 34800 and 36126 contain 6, 7 and 11 listed configurations. Ten omit an article number; their explicit catalogue IDs remain stable identifiers. All cards retain original decimal values, linear mass in kg/m and flat/trefoil buried-current conditions. Missing ratings remain null. Nexans confirms TSLF-O family use at Bessakerfjellet, but does not disclose installed cross-sections, article numbers or revisions. Family references are stored apart from exact installed-equipment links. These are OEM specifications, not independently validated performance. No model/voltage/section Cartesian product is generated.
The cables adapter and extended layouts adapter have synthetic offline fixtures.
The shared viewer exposes unavailable article numbers, unknown/zero linear mass,
and distinct current-rating conditions. The gallery compares listed cable geometry
without inferring installed length or total mass. Numerical curve coverage is unchanged.
See research/2026-11-11-morning-review.json for checksums, exact candidate rosters,
additions, enrichments, exclusions and quarantines. Next are BOE April 27–May 2,
NVE Hitra 19/Hitra2 68 and the Havøygavlen rating conflict. The 22 HARTING installed
revision checks and 13 original duplicate-name groups remain unresolved. Do not
repeat completed national inventories or the three cable-card partitions.
Hitra generations, Fosen fact cards and April BOE decisions#
The 2026-11-11-followup increment reuses PR #275 and its published morning
checkpoint. It adds five projects and enriches six, with no equipment, turbine
or curve additions. Branch totals are 8,480 projects, 657 equipment records,
13,679 turbine records, two numerical curves and 560 sources. Merged main at
db316472 contains 8,473 / 633 / 13,662 / 2 records and 532 sources.
The 50-project and 20-equipment objectives were not met in this increment.
BOE April 27–May 2 is a complete discovered A/B wind-title partition: ten notices reviewed, three additions (Crecente, Bustatur, Ampliación Tórtoles), one existing Mudarra enrichment, and six solar/battery-context exclusions. Crecente’s application refusal is dated and appeal outcomes remain unknown. Bustatur and Tórtoles retain earlier construction authorizations; land occupation and public-utility declarations do not establish construction or operation. No landowner names or landowner-level annex inventory is imported.
Mudarra’s April environmental decision describes 16 Siemens Gamesa turbines; the later October notice describes 15 Envision turbines. The earlier configuration, foundation, electrical design and source discrepancies remain dated observations. The current October model links, fleet, scalar capacity, status and creation time are preserved. Neither design is counted as a second project or installed fleet.
The saved NVE Hitra 19 and Hitra 2 68 records are reconciled with separate permit cases 10/34 and Statkraft/Fosen evidence. Hitra’s historical 55 MW is preserved alongside reported 55.2 MW. Its proposed replacement, Nye Hitra 1 (case 202501542), has a stable predecessor link and shared overlap group. Both generations remain excluded from totals. Hitra 2 is explicitly unaffected by the replacement and has independently reported 93.6 MW; 26 times 3.6 MW is only an arithmetic check. Hitra 2’s permit expiry is reported as 2044 in case 34 and 2045 in the replacement summary: no normalized expiry is asserted. The replacement’s 85–95 MW design range and up-to-120 MW application envelope remain separate, with no scalar MW.
All five Fosen operator fact cards were processed by the checksummed, roster-bound
wind_catalogue.fosen adapter: Storheia, Kvenndalsfjellet, Harbaksfjellet,
Geitfjellet and Hitra 2. Exact source model labels, rotor diameter, tower height,
construction/assembly/commissioning years, road lengths, plan areas and reported
annual output are retained. Tower height is not inferred hub height. Annual
output lacks a statistical period and gross/net basis; it is not converted to
measured generation or capacity factor. No exact installed equipment revision
or individual turbine identity is inferred. Storheia’s narrative handover in
February 2020 and its card’s 2019 commissioning year describe different milestones.
The shared viewer exposes missing periods/revisions and the deterministic gallery
compares reported geometry without aggregating generation.
The old Statkraft English Hitra 2 URL returned HTTP 404, not an authentication failure or explicit reuse restriction. Norwegian operator, Fosen and NVE pages were acquired with original-byte checksums. The dated review ledger records all captures, exclusions, additions, enrichments and remaining conflicts.
Next: BOE April 20–25 and saved NVE IDs 7 (Fjeldskår) and 31 (Smøla), without repeating national scans or completed Fosen cards. The 13 original duplicate-name groups and historical component-revision questions remain unresolved.
BOE April 20–25 and continuation identity checks#
The 2026-11-11 evening batch reviews every discovered A/B daily index for April 20–25. Ado (PEol-1019) remains an announced 68 MW proposal; its ten V162 positions are published in ETRS89 zone 29, not zone 30. The position adapter must explicitly support that table before individual records are added. Viento de Sellía (PE-227) has construction authorization and a dated public utility declaration, not evidence of operation. Its four 4.5 MW turbine ratings provide an 18 MW arithmetic check; the 16 MW evacuation limit is a separate boundary. No installed-capacity scalar is inferred from either.
Continuation labels are checked against saved registry observations using
resolve_saved_identity. NVE ID 7 is Hundhammerfjellet and ID 31 is Høg-Jæren;
the earlier continuation labels were incorrect. Fjeldskår is ID 1 and Smøla
is ID 5. The resolver rejects missing, duplicate and mismatched identities
without changing the existing review decision or aggregation exclusions.
Smøla’s 145.8 MW registry observation, 150.4 MW permit and 150 MW proposed
replacement require separate generation and date boundaries; none are summed.
See the dated research review for original-byte hashes, source locators, excluded personal annex data and exact remaining tasks.
2026-11-13: BOE April13–18 and Ado zone29 layout#
All six daily indexes and every discovered A/B wind-title notice were reviewed. Hedreira PEol-914 is a withdrawn application; its historical75MW is not installed or current planned capacity. Satrebares retains its later cancellation while the initial29-turbine/124.98MW and final18-turbine/88.63MW rejected designs remain dated observations. Four solar/storage hybridization notices remain held for wind identity crosswalks; their capacities are not wind additions.
Ado’s complete AD-01–AD-10 table in BOE-B-2026-12797 section1.2 supplies ten linked proposed positions. The adapter preserves ETRS89/UTM29N (EPSG:25829), original Spanish coordinate strings and one-metre published resolution. Positional accuracy, geographic coordinates and installed-model revision remain unknown. Mast and boundary vertices are excluded; unrelated merged budget cells cannot invalidate a correctly identified position table, while merged cells within that table still fail review. The shared viewer and deterministic gallery distinguish resolution from accuracy.
Replay the reviewed original, retained outside Git, with:
uv run python scripts/stage_wind_layout.py /path/to/ado.html \
catalogues/wind/research/2026-11-13-morning-ado-layout.json /tmp/ado.jsonl \
--now 2026-11-13T08:25:00Z
The checksum must match the review; output creation is exclusive. NVE7/31 followup pages clarify repowering and commissioning cohorts but original permit boundaries remain held. See the dated review ledger for exact next document URLs. No equipment record was accepted in this batch;50project/20equipment targets were not met.